Letter

Conditional Release and Reduced Mortality

To the Editor: If the purpose of commentaries is to demonstrate the irrationality of consumer advocates, the one in the November 2006 issue accomplishes this purpose well. In Taking Issue, Bernstein ( 1 ) commented on a study by Segal and Burgess ( 2 ). These authors examined case register data for several thousand patients in Australia and found that over a 13.5-year period those who had been given conditional release from a psychiatric hospitalization experienced a 14 percent reduction in the probability of noninjury-related death and a 24 percent reduction in injury-related death. Bernstein states that the study's methods were insufficient to establish causal links showing that conditional release exercised a protective effect in terms of subsequent mortality.

Does anyone believe that if an intervention were to increase mortality by similar rates, advocates would dismiss it with the comment that association does not establish causality? No—they would be shouting from the rooftops that so strong an association establishes causality beyond any doubt.

Moreover, what is the basis for the statements that public systems in Australia are "neglectful" of patients and that 15-minute outpatient appointments are inadequate? Judging from this large-scale study, it would seem that they are adequate to the task of these large reductions in mortality. Bernstein dismisses these reductions as "blunt indicators of clinical effectiveness," when, in truth, they are impossible-to-manipulate proxy measures, unlike some of the presumably more sophisticated measurements that are the dream of spin doctors everywhere.